ElevenLabs says its audio watermark detects only its own generated speech
Overview
ElevenLabs says its audio watermarking tool can give a definitive yes or no on whether audio was generated by its own products.
The company says conventional audio deepfake detection gives only a probability score. It also says watermark detection works only on audio from ElevenLabs, not on voice samples from other providers, so the tool should be one part of a broader deepfake-detection approach.
Written by AI from the articles below · updated Oct 14, 7:47 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- ElevenLabs BlogOfficialElevenLabs explains how to detect audio deepfakes with watermarks
AIElevenLabs says conventional audio deepfake detection gives only a probability score, while its watermarking tool returns a definitive yes or no for audio its own products generated. The company says watermark detection works only on its own audio, not on voice samples from other providers, so it should be one tool in a broader deepfake-detection stack.
Heat trend
Not enough continuous observations to show a trend yet.